one-way doors and two-way doors
the most common decision mistake is not deciding wrongly. it is spending one-way-door care on a two-way-door call, and the reverse.
jeff bezos set this out in the 2015 amazon shareholder letter, and it has survived because it takes ten seconds and resolves an argument you would otherwise have with yourself for a week.
some decisions are doors you can walk back through. some are not. they deserve completely different amounts of care, and treating them the same is the actual error.
two-way doors
you can undo it. a pricing test, a new channel, a tool, a job title, a landing page, a supplier trial.
the right speed here is fast. decide with roughly seventy percent of the information, watch what happens, reverse if it is wrong. the cost of being wrong is a few weeks. the cost of being slow is every week you were not learning.
the failure mode is obvious once named: applying board-level deliberation to something you could simply have tried.
one-way doors
you cannot easily undo it. a co-founder, a funding round, letting someone go, a rebrand, a platform migration, a long lease.
here, slow is correct. get the extra information. run a pre-mortem. talk to someone who has done it. the asymmetry justifies the time.
but still set a date. taking care is not the same as taking forever, and one-way doors are exactly where "i am still thinking about it" quietly turns into eighteen months.
the honest complication
the model is cleaner than reality. most decisions are somewhere in between, and a few are two-way doors that quietly become one-way if you leave them long enough.
the useful version of the question is not "is this reversible" but "if this is wrong, what does it cost me to undo, in money and in time?" answer that in one sentence. if the answer is small, move now. if you cannot answer it at all, that is your missing information and you have just found your next step.
a rough sorting test
- could i reverse this within a quarter for less than a month's revenue? two-way door. decide this week.
- would reversing it mean an awkward public correction? treat it as one-way even if it is technically reversible. reputational cost is real cost.
- does it involve a person's employment or equity? one-way. always.
- am i unable to tell? write the undo cost in one sentence. the sentence will tell you.
why founders get this backwards
because the anxiety a decision produces has almost nothing to do with how reversible it is. it tracks how visible it is, and how much of your identity is attached to it.
a pricing change is trivially reversible and terrifying. a two-year contract is nearly permanent and gets signed without much thought because it arrives as admin.
the door test is useful precisely because it ignores how the decision feels and asks only what undoing it costs.
if one of these has been open a while, the decision fatigue calculator puts a number on the waiting. and if you know it is a two-way door and still cannot move, the problem is not the door.